Whilst Open Source has significantly increased the quantity and quality of available building blocks, software development process itself is not undergoing more commoditization now than 10 years ago.
A programmer nowadays has more a less the same tools and uses more a less the same methodologies available by 1990.
Still one key diffence between software and other technologies is that if your have a couple of television channels
and you merge them, the result is a single television channel. But if you take a couple of software applications
and merge them the result is something that simply doesn't work unless you almost entirely rewrite one of the applications.
Another key problem of software development is that still the best way of developing a program is inside one person's head,
or at least inside the fewest people heads. That is why best core development teams always try to keep themselves small.

THERE ARE A LOT MORE OPEN SOURCE COMPONENTS NOW, BUT WE STILL DEVELOP SOFTWARE MUCH THE SAME AS WE DID 10 OR 20 YEARS AGO.

About Creative Destruction:

"Creative Destruction" is all right. What it is not so all right is "complete market destruction".
The key point here is not how Open Source changes a market such as Windows vs. Linux but how Open Source changes the whole production and benefits share.
Let me put an example: Ten years ago being an Oracle DBA was a real value. Oracle was so expensive (at least in my country) that even private universities could not afford buying it for educational purposes. As a result, students came out of the university with no idea of Oracle, and since Oracle was so common among big enterprises these people had to pay for training and then got higher salaries for their unique knowlegde. Today anybody anywhere around the world can download MySQL or PostgreSQL and learn everything about them (much more they could learn about Oracle). It is a matter of time that an Open Source database reach the performance
and reliability level of a good commercial database. It doesn't matter whether Oracle or PostgreSQL or anybody else wins. What does matter is that Open Source is moving the power of innovation from more developed countries to less developed countries and this is extremely important because the whole technology wealth of the last 200 years from fireguns to microchips has been based on selling knowledge and patented materials to people who just could not do it themselves. This is happening only in the
software industry and not is the hardware industry which is everyday more and more concentrated in a few points. One reasonable argument against extreme Open Source is that we'd better think twice where is our money going to come from before giving away our competive advantages. Change and paradigm shifts are all right, bloody revolutions are not.

OPEN SOURCE VS. PROPIETARY IS NOT JUST A MARKET WAR BUT A WORLDWIDE REDISTRIBUTION OF KNOWLEDGE AND POWER.

About building custom distributions:

A custom distribution is very expensive to mantain. Maybe you can do a fork of some well know code and sell it to government agency for a very specific purpose. But basing a business in custom distribution is suicide because there are so few people with enought money to pay for it.
Customizing software is not like painting a car in different colors.

ONE TO ONE CUSTOMIZATION MARKETING IS STILL NOT FEASIBLE BECAUSE SINGLE CUSTOM SMALL PIECES OF SOFTWARE ARE TOO EXPENSIVE TO PRODUCE.

About building a strong brand:

Building a strong brand with nothing behind it is one of the best ways of burning cash.
The business history is full of example of companies that spent their money in branding because thay did not know what else to advertise.

BRANDING IS LIKE COMMON SENSE: YOU SAY SOMETHING IS COMMON SENSE WHEN YOU CANNOT DEMOSTRATE IT.

About Network Collaboration and the Bazaar Model:

Collaborative Software development is quantum leap in the way software is build,
but it is not the ultimate solution to every problem. One basic thing to understand in
collaborative software development is that people develop things they need for themselves and then
put their work into public domain when they cannot make money out of it (this is how Stallman began writting his own printer drivers). As a consecuence it is difficult to control the output of a collaborative process. And if there is something that customers dislike that is uncertainity. Another barrier for Collaborative Software development is that we still lack of network enabled tools powerfull enougth
for supporting it. Publishing your work at SourceForge and doing some bug tracking is one thing, coordinating a hundred people
around the world writting ten thousand features is another one much more complicated.

COLLABORATIVE SOFTWARE DEVELOPMENT IS A PARADIGM SHIFT BUT NOT THE ULTIMATE SOLUTION TO SOFTWARE BUILDING.

Heroes of code fork:
I really love the short story about Mark Anders and Scott Guthrie ASP.NET prototype.
It demostrates that it is not just an idea but the energy you put on it what makes it work.

A SINGLE VISION CAN CHANGE THE WORLD IF YOU PERSIST ENOUGHT ON IT.